By The Carpenter on December 07, 2009

Numbers can lie

Smart statisticians, economists, CEOs can play with numbers to make them look good.

That is why companies play with their ROI models changing them whenever they feel they should. That is why executives spend a great deal of time choosing their words in the opening statements of the annual reports.

I got the idea for this post when I was reading InBusiness the other day and there was a section saying that the average monthly salary for men in Cyprus is €1,833 and for women about €1,580. I have always claimed that the “average” is not always the best type of measurement and in my humble opinion it is very wrong to use this for the average wage.
Let me explain myself:

In recording all wages, there are many outliers. Probably an average job in Cyprus would pay between €1100 to €2000. But there are many people who earn much more than that. Those outliers are always on the top side and not the bottom. We cannot have outliers on the bottom because according to regulations there is a minimum wage that needs to be paid. So for every person in the population the lowest wage you can find equals the minimum wage. On the other hand, there can be many outliers on the top side. There are people who earn € 5,000 a month, there are people who earn €10,000 and there are people who earn €30,000 a month. These figures unavoidably inflate the average salary giving a misrepresentation. All things considered I believe that the use of the median would probably be much more representative in cases of salaries.

Thinking about numbers, I also remembered an article I read about Lawrence Summers, ex-President of Harvard. If you ask pretty much anyone they would tell you that Mr. Summers claimed in a business lunch that “men are smarter than women”. What he actually said is that “there is greater variance in the intelligence of men than women”. Of course Mr. Summers is not stupid, he is actually very smart and he made his comment after years of research. Greater variance means that the tails of a normal curve of the IQ of men are a bit larger than the respective ones of the women. There might be more men than women who demonstrate a higher than normal IQ , but there are also more men than women who demonstrate lower than normal IQ.

1 comment:

  1. The final GDP figures of an OECD country are finalized 5 years after the original publication. The error is +/- 1.5% rising to +/- 2.5% in some countries that have poorer data availability. In quarterly GDP that rises significantly.

    So imagine my surprise to hear about "coming out of a downturn" when quarterly GDP in the UK was 0.1%!!!!

    ReplyDelete